What are Neural Voices?

Neural voices are advanced text-to-speech (TTS) options offered by providers like AWS, Google, and Microsoft. These voices use AI-based language models to produce speech that is more natural and human-like compared to standard voices.

How to Use Neural Voices

Neural voices are not included in the “Authoring” package due to their higher costs. To use neural voices in a project, it must be created with a “Production” account. Once the project is created, all invited team members can also use the neural voices, even if they only have an “Authoring” subscription.

Important: Make sure that the person with the “Production” account creates the project!

How to Identify a Neural Voice

Neural voices stand out due to fewer pronunciation errors and superior sound quality. In addition, each provider uses specific naming conventions for these voices:

AWS

For AWS Polly, maintaining an up-to-date list of neural voices is challenging, as they frequently update or convert standard voices into neural ones. Check the official list of AWS neurals. If a voice in Frazier isn’t on this list, it’s likely a standard voice.

Google

Google offers different quality levels for its voices.

Standard Voices

All standard voices can be identified by their name: “de-DE-Standard-*”. The asterisk is replaced by a letter, currently ranging from A to F.

Neural Voices

You can also recognize neural voices by their name. It contains the words “Wavenet”, “Neural2”, or “Polyglot”.

Tip: Google’s naming system allows for seamless transitions. For example, “de-DE-Standard-B” can be upgraded to “de-DE-Wavenet-B” with minimal adjustments, as both represent the same speaker’s recordings..

Microsoft

All Microsoft voices available now are neural voices.

Last Update: 2026/01/23